Classifications

• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L15/00—Speech recognition
  • G10L15/08—Speech classification or search
  • G10L15/16—Speech classification or search using artificial neural networks
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
  • G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
  • G10L21/038—Speech enhancement, e.g. noise reduction or echo cancellation using band spreading techniques
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N3/00—Computing arrangements based on biological models
  • G06N3/02—Neural networks
  • G06N3/08—Learning methods
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N3/00—Computing arrangements based on biological models
  • G06N3/02—Neural networks
  • G06N3/08—Learning methods
  • G06N3/084—Backpropagation, e.g. using gradient descent
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L15/00—Speech recognition
  • G10L15/02—Feature extraction for speech recognition; Selection of recognition unit
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L15/00—Speech recognition
  • G10L15/06—Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice
  • G10L15/063—Training
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
  • G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
  • G10L25/18—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters the extracted parameters being spectral information of each sub-band
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
  • G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
  • G10L25/21—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters the extracted parameters being power information
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
  • G10L25/93—Discriminating between voiced and unvoiced parts of speech signals
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N3/00—Computing arrangements based on biological models
  • G06N3/02—Neural networks
  • G06N3/04—Architecture, e.g. interconnection topology
  • G06N3/045—Combinations of networks
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L15/00—Speech recognition
  • G10L15/02—Feature extraction for speech recognition; Selection of recognition unit
  • G10L2015/025—Phonemes, fenemes or fenones being the recognition units
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
  • G10L25/93—Discriminating between voiced and unvoiced parts of speech signals
  • G10L2025/937—Signal energy in various frequency bands
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
  • G10L25/27—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the analysis technique
  • G10L25/30—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the analysis technique using neural networks

Definitions

• UMTS Universal Mobile Telecommunications System
• LTE Long-Term Evolution
• Coverage tends to be low for locations such as remote highways or rural areas in the countryside.
• narrow-band e.g., 8kHz
• Bandwidth extension may be used to bridge the arising voice quality gap by artificially extending the narrow-band (e.g., 8kHz) telephone signal to a wideband (e.g., l6kHz), super-wideband (e.g., 24kHz) or even full-band (e.g., 32/48kHz) signal.
• narrow-band e.g. 8kHz
• wideband e.g., l6kHz
• super-wideband e.g., 24kHz
• full-band e.g., 32/48kHz
• a method, performed by one or more computing devices may include but is not limited to transforming, by a computing device, a speech signal into a speech signal representation.
• a regression deep neural network may be trained with a cost function to minimize a mean squared error between actual values of the speech signal representation and estimated values of the speech signal representation, wherein the cost function may include one or more discriminative terms.
• the bandwidth of the speech signal may be extended by extending the speech signal representation of the speech signal using the regression deep neural network trained with the cost function that includes the one or more discriminative terms.
• the speech signal representation may be obtained by decomposing the speech signal into a spectral envelope and an excitation signal, and wherein the spectral envelope may be extended using the regression deep neural network trained with the cost function.
• the one or more discriminative terms may include at least one of a fricative-to-vowel power ratio and a function thereof.
• the one or more discriminative terms may preserve relations of statistics between different phoneme classes in the actual values of the speech signal representation and the estimated values of the speech signal representation.
• the cost function may preserve a power ratio between the different phoneme classes in the actual values of the speech signal representation and the estimated values of the speech signal representation.
• the cost function may preserve the power ratio between the different phoneme classes using a weighted sum of K power ratio errors between the different phoneme classes. An average power ratio may be reproduced at an output of the regression deep neural network.
• a computing system may include one or more processors and one or more memories configured to perform operations that may include but are not limited to transforming a speech signal into a speech signal representation.
• a regression deep neural network may be trained with a cost function to minimize a mean squared error between actual values of the speech signal representation and estimated values of the speech signal representation, wherein the cost function may include one or more discriminative terms.
• the bandwidth of the speech signal may be extended by extending the speech signal representation of the speech signal using the regression deep neural network trained with the cost function that includes the one or more discriminative terms.
• the speech signal representation may be obtained by decomposing the speech signal into a spectral envelope and an excitation signal, and wherein the spectral envelope may be extended using the regression deep neural network trained with the cost function.
• the one or more discriminative terms may include at least one of a fricative-to-vowel power ratio and a function thereof.
• the one or more discriminative terms may preserve relations of statistics between different phoneme classes in the actual values of the speech signal representation and the estimated values of the speech signal representation.
• the cost function may preserve a power ratio between the different phoneme classes in the actual values of the speech signal representation and the estimated values of the speech signal representation.
• the cost function may preserve the power ratio between the different phoneme classes using a weighted sum of K power ratio errors between the different phoneme classes. An average power ratio may be reproduced at an output of the regression deep neural network.
• a computer program product may reside on a computer readable storage medium having a plurality of instructions stored thereon which, when executed across one or more processors, may cause at least a portion of the one or more processors to perform operations that may include but are not limited to transforming a speech signal into a speech signal representation.
• a regression deep neural network may be trained with a cost function to minimize a mean squared error between actual values of the speech signal representation and estimated values of the speech signal representation, wherein the cost function may include one or more discriminative terms.
• the bandwidth of the speech signal may be extended by extending the speech signal representation of the speech signal using the regression deep neural network trained with the cost function that includes the one or more discriminative terms.
• the speech signal representation may be obtained by decomposing the speech signal into a spectral envelope and an excitation signal, and wherein the spectral envelope may be extended using the regression deep neural network trained with the cost function.
• the one or more discriminative terms may include at least one of a fricative-to-vowel power ratio and a function thereof.
• the one or more discriminative terms may preserve relations of statistics between different phoneme classes in the actual values of the speech signal representation and the estimated values of the speech signal representation.
• the cost function may preserve a power ratio between the different phoneme classes in the actual values of the speech signal representation and the estimated values of the speech signal representation.
• the cost function may preserve the power ratio between the different phoneme classes using a weighted sum of K power ratio errors between the different phoneme classes. An average power ratio may be reproduced at an output of the regression deep neural network.
• Fig. 1 is an example diagrammatic view of a training process coupled to an example distributed computing network according to one or more example implementations of the disclosure
• Fig. 2 is an example diagrammatic view of a computer and client electronic device of Fig. 1 according to one or more example implementations of the disclosure;
• Fig. 3 is an example diagrammatic view of a source/filter model according to one or more example implementations of the disclosure
• Fig. 4 is an example diagrammatic view of a bandwidth extension architecture according to one or more example implementations of the disclosure.
• Fig. 5 is an example diagrammatic view of a spectral envelope, an example excitation and the synthesized spectrum according to one or more example implementations of the disclosure
• Fig. 6 is an example diagrammatic view of a deep-neural-network-based bandwidth extension system according to one or more example implementations of the disclosure
• Fig. 7 is an example diagrammatic view of a feedforward neural network according to one or more example implementations of the disclosure.
• Fig. 8 is an example diagrammatic view of non-linear activation functions according to one or more example implementations of the disclosure.
• Fig. 9 is an example diagrammatic view of the calculation of activation energies in feedforward neural networks according to one or more example implementations of the disclosure.
• Fig. 10 is an example diagrammatic view of a bandwidth-extended speech spectrogram with MSE cost function and a true wideband speech spectrogram according to one or more example implementations of the disclosure;
• Fig. 11 is an example diagrammatic view of a plot according to one or more example implementations of the disclosure.
• Fig. 12 is an example flowchart of a training process according to one or more example implementations of the disclosure.
• Fig. 13 is an example diagrammatic view of plots according to one or more example implementations of the disclosure.
• the present disclosure may be embodied as a method, system, or computer program product. Accordingly, in some implementations, the present disclosure may take the form of an entirely hardware implementation, an entirely software implementation (including firmware, resident software, micro-code, etc.) or an implementation combining software and hardware aspects that may all generally be referred to herein as a“circuit,”“module” or“system.” Furthermore, in some implementations, the present disclosure may take the form of a computer program product on a computer-usable storage medium having computer-usable program code embodied in the medium.
• any suitable computer usable or computer readable medium may be utilized.
• the computer readable medium may be a computer readable signal medium or a computer readable storage medium.
• the computer-usable, or computer-readable, storage medium (including a storage device associated with a computing device or client electronic device) may be, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, device, or any suitable combination of the foregoing.
• the computer-readable medium may include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a digital versatile disk (DVD), a static random access memory (SRAM), a memory stick, a floppy disk, a mechanically encoded device such as punch-cards or raised structures in a groove having instructions recorded thereon, a media such as those supporting the internet or an intranet, or a magnetic storage device.
• RAM random access memory
• ROM read-only memory
• EPROM or Flash memory erasable programmable read-only memory
• CD-ROM compact disc read-only memory
• DVD digital versatile disk
• SRAM static random access memory
• a memory stick a floppy disk
• a mechanically encoded device such as punch-cards or
• a computer-usable or computer-readable, storage medium may be any tangible medium that can contain or store a program for use by or in connection with the instruction execution system, apparatus, or device.
• a computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. In some implementations, such a propagated signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof.
• the computer readable program code may be transmitted using any appropriate medium, including but not limited to the internet, wireline, optical fiber cable, RF, etc.
• a computer readable signal medium may be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
• computer program code for carrying out operations of the present disclosure may be assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state-setting data, or either source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Java ® , Smalltalk, C++ or the like. Java ® and all Java- based trademarks and logos are trademarks or registered trademarks of Oracle and/or its affiliates.
• the computer program code for carrying out operations of the present disclosure may also be written in conventional procedural programming languages, such as the "C" programming language, PASCAL, or similar programming languages, as well as in scripting languages such as Javascript, PERL, or Python.
• the program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user’s computer and partly on a remote computer or entirely on the remote computer or server.
• the remote computer may be connected to the user’s computer through a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (for example, through the internet using an Internet Service Provider).
• LAN local area network
• WAN wide area network
• Internet Service Provider for example, AT&T, MCI, Sprint, EarthLink, etc.
• electronic circuitry including, for example, programmable logic circuitry, field-programmable gate arrays (FPGAs) or other hardware accelerators, micro- controller units (MCUs), or programmable logic arrays (PLAs) may execute the computer readable program instructions/code by utilizing state information of the computer readable program instructions to personalize the electronic circuitry, in order to perform aspects of the present disclosure.
• FPGAs field-programmable gate arrays
• MCUs micro- controller units
• PDAs programmable logic arrays
• the flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of apparatus (systems), methods and computer program products according to various implementations of the present disclosure.
• Each block in the flowchart and/or block diagrams, and combinations of blocks in the flowchart and/or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable computer program instructions for implementing the specified logical function(s)/act(s).
• These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the computer program instructions, which may execute via the processor of the computer or other programmable data processing apparatus, create the ability to implement one or more of the functions/acts specified in the flowchart and/or block diagram block or blocks or combinations thereof.
• the functions noted in the block(s) may occur out of the order noted in the figures (or combined or omitted).
• two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved.
• these computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function/act specified in the flowchart and/or block diagram block or blocks or combinations thereof.
• the computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed (not necessarily in a particular order) on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions/acts (not necessarily in a particular order) specified in the flowchart and/or block diagram block or blocks or combinations thereof.
• training process 10 may reside on and may be executed by a computer (e.g., computer 12), which may be connected to a network (e.g., network 14) (e.g., the internet or a local area network).
• a network e.g., network 14
• client electronic devices may include, but are not limited to, a storage system (e.g., a Network Attached Storage (NAS) system, a Storage Area Network (SAN)), a personal computer(s), a laptop computer(s), mobile computing device(s), a server computer, a series of server computers, a mainframe computer(s), or a computing cloud(s).
• NAS Network Attached Storage
• SAN Storage Area Network
• a SAN may include one or more of the client electronic devices, including a RAID device and a NAS system.
• each of the aforementioned may be generally described as a computing device.
• a computing device may be a physical or virtual device.
• a computing device may be any device capable of performing operations, such as a dedicated processor, a portion of a processor, a virtual processor, a portion of a virtual processor, portion of a virtual device, or a virtual device.
• a processor may be a physical processor or a virtual processor.
• a virtual processor may correspond to one or more parts of one or more physical processors.
• the instructions/logic may be distributed and executed across one or more processors, virtual or physical, to execute the instructions/logic.
• Computer 12 may execute an operating system, for example, but not limited to, Microsoft® Windows®; Mac® OS X®; Red Hat® Linux®, Windows® Mobile, Chrome OS, Blackberry OS, Fire OS, or a custom operating system.
• Microsoft and Windows are registered trademarks of Microsoft Corporation in the United States, other countries or both
• Mac and OS X are registered trademarks of Apple Inc. in the United States, other countries or both
• Red Hat is a registered trademark of Red Hat Corporation in the United States, other countries or both
• Linux is a registered trademark of Linus Torvalds in the United States, other countries or both).
• a training process such as training process 10 of Fig. 1, may transform, by a computing device, a speech signal into a speech signal representation.
• a regression deep neural network may be trained with a cost function to minimize a mean squared error between actual values of the speech signal representation and estimated values of the speech signal representation, wherein the cost function may include one or more discriminative terms.
• Bandwidth of the speech signal may be extended by extending the speech signal representation of the speech signal using the regression deep neural network trained with the cost function that includes the one or more discriminative terms.
• the instruction sets and subroutines of training process 10 may be stored on storage device, such as storage device 16, coupled to computer 12, may be executed by one or more processors and one or more memory architectures included within computer 12.
• storage device 16 may include but is not limited to: a hard disk drive; all forms of flash memory storage devices; a tape drive; an optical drive; a RAID array (or other array); a random access memory (RAM); a read-only memory (ROM); or combination thereof.
• storage device 16 may be organized as an extent, an extent pool, a RAID extent (e.g., an example 4D+1P R5, where the RAID extent may include, e.g., five storage device extents that may be allocated from, e.g., five different storage devices), a mapped RAID (e.g., a collection of RAID extents), or combination thereof.
• a RAID extent e.g., an example 4D+1P R5, where the RAID extent may include, e.g., five storage device extents that may be allocated from, e.g., five different storage devices
• a mapped RAID e.g., a collection of RAID extents
• network 14 may be connected to one or more secondary networks (e.g., network 18), examples of which may include but are not limited to: a local area network; a wide area network; or an intranet, for example.
• secondary networks e.g., network 18
• networks may include but are not limited to: a local area network; a wide area network; or an intranet, for example.
• computer 12 may include a data store, such as a database (e.g., relational database, object-oriented database, triplestore database, etc.) and may be located within any suitable memory location, such as storage device 16 coupled to computer 12.
• a database e.g., relational database, object-oriented database, triplestore database, etc.
• data, metadata, information, etc. described throughout the present disclosure may be stored in the data store.
• computer 12 may utilize any known database management system such as, but not limited to, DB2, in order to provide multi-user access to one or more databases, such as the above noted relational database.
• the data store may also be a custom database, such as, for example, a flat file database or an XML database.
• training process 10 may be a component of the data store, a standalone application that interfaces with the above noted data store and/or an applet / application that is accessed via client applications 22, 24, 26, 28.
• the above noted data store may be, in whole or in part, distributed in a cloud computing topology.
• computer 12 and storage device 16 may refer to multiple devices, which may also be distributed throughout the network.
• computer 12 may execute a speech recognition application (e.g., speech recognition application 20), examples of which may include, but are not limited to, e.g., an automatic speech recognition application, a video conferencing application, a voice-over-IP application, a video-over-IP application, an Instant Messaging (IM)/"chat” application, a short messaging service (SMS)/multimedia messaging service (MMS) application, a telephony network application, a Speech Signal Enhancement (SSE) application, or other application that allows for virtual meeting and/or remote collaboration and/or recognition/translation of spoken language into text (and vice versa) by computing devices.
• IM Instant Messaging
• MMS multimedia messaging service
• SSE Speech Signal Enhancement
• training process 10 and/or speech recognition application 20 may be accessed via one or more of client applications 22, 24, 26, 28.
• training process 10 may be a standalone application, or may be an applet / application / script / extension that may interact with and/or be executed within speech recognition application 20, a component of speech recognition application 20, and/or one or more of client applications 22, 24, 26, 28.
• speech recognition application 20 may be a standalone application, or may be an applet / application / script / extension that may interact with and/or be executed within training process 10, a component of training process 10, and/or one or more of client applications 22, 24, 26, 28.
• client applications 22, 24, 26, 28 may be a standalone application, or may be an applet / application / script / extension that may interact with and/or be executed within and/or be a component of training process 10 and/or speech recognition application 20.
• client applications 22, 24, 26, 28 may include, but are not limited to, e.g., an automatic speech recognition application, a video conferencing application, a voice-over- IP application, a video-over-IP application, an Instant Messaging (IM)/"chat” application, a short messaging service (SMS)/multimedia messaging service (MMS) application, a telephony network application, a Speech Signal Enhancement (SSE) application, or other application that allows for virtual meeting and/or remote collaboration and/or recognition/translation of spoken language into text (and vice versa) by computing devices, a standard and/or mobile web browser, an email application (e.g., an email client application), a textual and/or a graphical user interface, a customized web browser, a plugin, an Application Programming Interface (API), or a custom application.
• an automatic speech recognition application e.g., an email client application
• IM Instant Messaging
• MMS multimedia messaging service
• SE Speech Signal Enhancement
• the instruction sets and subroutines of client applications 22, 24, 26, 28, which may be stored on storage devices 30, 32, 34, 36, coupled to client electronic devices 38, 40, 42, 44, may be executed by one or more processors and one or more memory architectures incorporated into client electronic devices 38, 40, 42, 44.
• one or more of storage devices 30, 32, 34, 36 may include but are not limited to: hard disk drives; flash drives, tape drives; optical drives; RAID arrays; random access memories (RAM); and read-only memories (ROM).
• client electronic devices 38, 40, 42, 44 may include, but are not limited to, a personal computer (e.g., client electronic device 38), a laptop computer (e.g., client electronic device 40), a smart/data-enabled, cellular phone (e.g., client electronic device 42), a notebook computer (e.g., client electronic device 44), a tablet, a server, a television, a smart television, a media (e.g., video, photo, etc.) capturing device, and a dedicated network device.
• a personal computer e.g., client electronic device 38
• a laptop computer e.g., client electronic device 40
• a smart/data-enabled, cellular phone e.g., client electronic device 42
• notebook computer e.g., client
• Client electronic devices 38, 40, 42, 44 may each execute an operating system, examples of which may include but are not limited to, AndroidTM, Apple® iOS®, Mac® OS X®; Red Hat® Linux®, Windows® Mobile, Chrome OS, Blackberry OS, Fire OS, or a custom operating system.
• training process 10 may be a purely server-side application, a purely client-side application, or a hybrid server-side / client-side application that is cooperatively executed by one or more of client applications 22, 24, 26, 28 and/or training process 10.
• one or more of client applications 22, 24, 26, 28 may be configured to effectuate some or all of the functionality of speech recognition application 20 (and vice versa). Accordingly, in some implementations, speech recognition application 20 may be a purely server-side application, a purely client-side application, or a hybrid server-side / client-side application that is cooperatively executed by one or more of client applications 22, 24, 26, 28 and/or speech recognition application 20.
• client applications 22, 24, 26, 28, training process 10, and speech recognition application 20 taken singly or in any combination, may effectuate some or all of the same functionality, any description of effectuating such functionality via one or more of client applications 22, 24, 26, 28, training process 10, speech recognition application 20, or combination thereof, and any described interaction(s) between one or more of client applications 22, 24, 26, 28, training process 10, speech recognition application 20, or combination thereof to effectuate such functionality, should be taken as an example only and not to limit the scope of the disclosure.
• one or more of users 46, 48, 50, 52 may access computer 12 and training process 10 (e.g., using one or more of client electronic devices 38, 40, 42, 44) directly through network 14 or through secondary network 18. Further, computer 12 may be connected to network 14 through secondary network 18, as illustrated with phantom link line 54. Training process 10 may include one or more user interfaces, such as browsers and textual or graphical user interfaces, through which users 46, 48, 50, 52 may access training process 10.
• the various client electronic devices may be directly or indirectly coupled to network 14 (or network 18).
• client electronic device 38 is shown directly coupled to network 14 via a hardwired network connection.
• client electronic device 44 is shown directly coupled to network 18 via a hardwired network connection.
• Client electronic device 40 is shown wirelessly coupled to network 14 via wireless communication channel 56 established between client electronic device 40 and wireless access point (i.e., WAP) 58, which is shown directly coupled to network 14.
• WAP 58 may be, for example, an IEEE 802.1 la, 802.1 lb, 802.
• Client electronic device 42 is shown wirelessly coupled to network 14 via wireless communication channel 60 established between client electronic device 42 and cellular network / bridge 62, which is shown by example directly coupled to network 14.
• some or all of the IEEE 802.1 lx specifications may use Ethernet protocol and carrier sense multiple access with collision avoidance (i.e., CSMA/CA) for path sharing.
• the various 802.1 lx specifications may use phase-shift keying (i.e., PSK) modulation or complementary code keying (i.e., CCK) modulation, for example.
• PSK phase-shift keying
• CCK complementary code keying
• BluetoothTM including BluetoothTM Low Energy
• NFC Near Field Communication
• FIG. 2 there is shown a diagrammatic view of computer 12 and client electronic device 42. While client electronic device 42 and computer 12 are shown in this figure, this is for example purposes only and is not intended to be a limitation of this disclosure, as other configurations are possible. Additionally, any computing device capable of executing, in whole or in part, training process 10 may be substituted for client electronic device 42 and computer 12 (in whole or in part) within Fig. 2, examples of which may include but are not limited to one or more of client electronic devices 38, 40, and 44.
• Client electronic device 42 and/or computer 12 may also include other devices, such as televisions with one or more processors embedded therein or attached thereto as well as any of the microphones, microphone arrays, and/or speakers described herein.
• the components shown here, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the disclosure described.
• computer 12 may include processor 202, memory 204, storage device 206, a high-speed interface 208 connecting to memory 204 and high speed expansion ports 210, and low speed interface 212 connecting to low speed bus 214 and storage device 206.
• processor 202 can process instructions for execution within the computer 12, including instructions stored in the memory 204 or on the storage device 206 to display graphical information for a GUI on an external input/output device, such as display 216 coupled to high speed interface 208.
• multiple processors and/or multiple buses may be used, as appropriate, along with multiple memories and types of memory.
• multiple computing devices may be connected, with each device providing portions of the necessary operations (e.g., as a server bank, a group of blade servers, or a multi-processor system).
• Memory 204 may store information within the computer 12.
• memory 204 may be a volatile memory unit or units.
• memory 204 may be a non-volatile memory unit or units.
• the memory 204 may also be another form of computer-readable medium, such as a magnetic or optical disk.
• Storage device 206 may be capable of providing mass storage for computer 12.
• the storage device 206 may be or contain a computer-readable medium, such as a floppy disk device, a hard disk device, an optical disk device, or a tape device, a flash memory or other similar solid state memory device, or an array of devices, including devices in a storage area network or other configurations.
• a computer program product can be tangibly embodied in an information carrier.
• the computer program product may also contain instructions that, when executed, perform one or more methods, such as those described above.
• the information carrier is a computer- or machine- readable medium, such as the memory 204, the storage device 206, memory on processor 202, or a propagated signal.
• High speed controller 208 may manage bandwidth-intensive operations for computer 12, while the low speed controller 212 may manage lower bandwidth-intensive operations. Such allocation of functions is exemplary only.
• the high-speed controller 208 may be coupled to memory 204, display 216 (e.g., through a graphics processor or accelerator), and to high-speed expansion ports 210, which may accept various expansion cards (not shown).
• low-speed controller 212 is coupled to storage device 206 and low-speed expansion port 214.
• the low-speed expansion port which may include various communication ports (e.g., USB, Bluetooth, Ethernet, wireless Ethernet) may be coupled to one or more input/output devices, such as a keyboard, a pointing device, a scanner, or a networking device such as a switch or router, e.g., through a network adapter.
• input/output devices such as a keyboard, a pointing device, a scanner, or a networking device such as a switch or router, e.g., through a network adapter.
• Computer 12 may be implemented in a number of different forms, as shown in the figure.
• computer 12 may be implemented as a standard server 220, or multiple times in a group of such servers. It may also be implemented as part of a rack server system 224.
• components from computer 12 may be combined with other components in a mobile device (not shown), such as client electronic device 42.
• client electronic device 42 Each of such devices may contain one or more of computer 12, client electronic device 42, and an entire system may be made up of multiple computing devices communicating with each other.
• Client electronic device 42 may include processor 226, memory 204, an input/output device such as display 216, a communication interface 262, and a transceiver 264, among other components.
• Client electronic device 42 may also be provided with a storage device, such as a microdrive or other device, to provide additional storage.
• a storage device such as a microdrive or other device, to provide additional storage.
• Each of the components 226, 204, 216, 262, and 264 may be interconnected using various buses, and several of the components may be mounted on a common motherboard or in other manners as appropriate.
• Processor 226 may execute instructions within client electronic device 42, including instructions stored in the memory 204.
• the processor may be implemented as a chipset of chips that include separate and multiple analog and digital processors.
• the processor may provide, for example, for coordination of the other components of client electronic device 42, such as control of user interfaces, applications run by client electronic device 42, and wireless communication by client electronic device 42.
• processor 226 may communicate with a user through control interface 258 and display interface 260 coupled to a display 216.
• the display 216 may be, for example, a TFT LCD (Thin-Film-Transistor Liquid Crystal Display) or an OLED (Organic Light Emitting Diode) display, or other appropriate display technology.
• the display interface 260 may comprise appropriate circuitry for driving the display 216 to present graphical and other information to a user.
• the control interface 258 may receive commands from a user and convert them for submission to the processor 226.
• an external interface 262 may be provide in communication with processor 226, so as to enable near area communication of client electronic device 42 with other devices. External interface 262 may provide, for example, for wired communication in some implementations, or for wireless communication in other implementations, and multiple interfaces may also be used.
• memory 204 may store information within the Client electronic device 42.
• the memory 204 can be implemented as one or more of a computer-readable medium or media, a volatile memory unit or units, or a non-volatile memory unit or units.
• Expansion memory 264 may also be provided and connected to client electronic device 42 through expansion interface 266, which may include, for example, a SIMM (Single In Line Memory Module) card interface.
• SIMM Single In Line Memory Module
• expansion memory 264 may provide extra storage space for client electronic device 42, or may also store applications or other information for client electronic device 42.
• expansion memory 264 may include instructions to carry out or supplement the processes described above, and may include secure information also.
• expansion memory 264 may be provide as a security module for client electronic device 42, and may be programmed with instructions that permit secure use of client electronic device 42.
• secure applications may be provided via the SIMM cards, along with additional information, such as placing identifying information on the SIMM card in a non-hackable manner.
• the memory may include, for example, flash memory and/or NVRAM memory, as discussed below.
• a computer program product is tangibly embodied in an information carrier.
• the computer program product may contain instructions that, when executed, perform one or more methods, such as those described above.
• the information carrier may be a computer- or machine-readable medium, such as the memory 204, expansion memory 264, memory on processor 226, or a propagated signal that may be received, for example, over transceiver 264 or external interface 262.
• Client electronic device 42 may communicate wirelessly through communication interface 262, which may include digital signal processing circuitry where necessary.
• Communication interface 262 may provide for communications under various modes or protocols, such as GSM voice calls, SMS, EMS, or MMS speech recognition, CDMA, TDMA, PDC, WCDMA, CDMA2000, or GPRS, among others.
• Such communication may occur, for example, through radio-frequency transceiver 264.
• short-range communication may occur, such as using a Bluetooth, WiFi, or other such transceiver (not shown).
• GPS Global Positioning System
• Client electronic device 42 may also communicate audibly using audio codec 270, which may receive spoken information from a user and convert it to usable digital information. Audio codec 270 may likewise generate audible sound for a user, such as through a speaker, e.g., in a handset of client electronic device 42. Such sound may include sound from voice telephone calls, may include recorded sound (e.g., voice messages, music files, etc.) and may also include sound generated by applications operating on client electronic device 42.
• Audio codec 270 may receive spoken information from a user and convert it to usable digital information. Audio codec 270 may likewise generate audible sound for a user, such as through a speaker, e.g., in a handset of client electronic device 42. Such sound may include sound from voice telephone calls, may include recorded sound (e.g., voice messages, music files, etc.) and may also include sound generated by applications operating on client electronic device 42.
• Client electronic device 42 may be implemented in a number of different forms, as shown in the figure. For example, it may be implemented as a cellular telephone 280. It may also be implemented as part of a smartphone 282, personal digital assistant, remote control, or other similar mobile device.
• UMTS Universal Mobile Telecommunications System
• LTE Long-Term Evolution
• Coverage tends to be low for locations such as remote highways or rural areas in the countryside.
• narrow-band e.g., 8kHz
• Bandwidth extension may be used to bridge the arising voice quality gap by artificially extending the narrow-band (e.g., 8kHz) telephone signal to a wideband (e.g., l6kHz), super-wideband (e.g., 24kHz) or even full-band (e.g., 32/48kHz) signal.
• narrow-band e.g. 8kHz
• wideband e.g., l6kHz
• super-wideband e.g., 24kHz
• full-band e.g., 32/48kHz
• artificial bandwidth extension may reconstruct a l6kHz wide-band signal from a given 8kHz narrow-band signal. This is typically used in the context of telephony networks and may be achieved by decomposing the speech signal into its excitation and its spectral envelope, both of which may then be extended separately.
• State-of-the-art approaches typically use deep (regression) neural networks (DNNs) for this task.
• DNNs deep neural networks
• MSE mean squared error
• the present disclosure may avoid over-smoothing (among other technical problems) while extending narrow-band speech from Global System for Mobile communications (GSM) / Code Division Multiple Access (CDMA) connections (or other connections) to high-quality wideband speech by, e.g., augmenting or adding additional discriminative terms to the cost function (e.g., the MSE cost function) that explicitly force the DNN to maintain a good separability of different phoneme classes (e.g., fricatives and vowels). These terms may increase the cost or “punish” deviations of the average power ratio (APR) between different phoneme classes, which may force the DNN to reproduce the APR of the training data at the output of the DNN. As such, the present disclosure may result in a higher speech quality of the bandwidth-extended signal with more natural bandwidth-extended speech (e.g., due to better separation of fricatives and vowels).
• GSM Global System for Mobile communications
• CDMA Code Division Multiple Access
• training process 10 may at least help, e.g., to overcome an example and non-limiting problem necessarily rooted in and specifically arising in the realm of computer and/or telephone networks to improve existing technological processes associated with, e.g., artificial bandwidth extension and over- smoothing.
• the general aim of artificial bandwidth extension is to improve the voice quality of communication (e.g., phone calls) by extending narrow-band (e.g., 8kHz bandlimited) speech from, e.g., GSM/CDMA connections to high-quality wideband (e.g., l6kHz bandwidth) speech (or other data).
• narrow-band e.g., 8kHz bandlimited
• wideband e.g., l6kHz bandwidth
• BWE systems typically make use of the source / filter model of speech production, such as model 300 shown in example Fig. 3.
• Training process 10 may use this model to separate speech into a glottal excitation signal and resonance frequencies of the vocal tract.
• the glottal excitation signal typically either consists of an“impulse train”-like signal in the case of voiced speech (such as vowels: a, e, i, o, u) or consists of white noise in the case of unvoiced sounds (fricatives, such as f, s, z, sh, ch).
• the resonance frequencies of the vocal tract may define the spectral envelope, including the formant frequencies that specify in which parts of the spectrum the speech energy is concentrated. They may be modeled as a Finite Impulse Response (FIR) filter that is applied to the glottis excitation signal.
• FIR Finite Impulse Response
• the typical architecture of BWE systems may look like architecture 400 shown in example Fig. 4.
• the NB excitation signal may be obtained by training process 10 removing the spectral envelope. In the frequency domain, this may be achieved by dividing the NB speech spectrum by the estimated envelope, as indicated in Fig. 4.
• the envelope and excitation may be extended separately by training process 10.
• the excitation may be often extended with methods such as spectral folding (e.g., mirroring the NB excitation upwards to frequencies above 4 kHz) or modulation (e.g., shifting the NB excitation upwards to frequencies above 4 kHz), a lot of effort is put into getting the extension of the spectral envelope correct. This is motivated by the fact that errors in the envelope extension typically result in much stronger artifacts of the bandwidth-extended speech signal than errors in the excitation.
• the estimated, e.g., bandwidth-extended wideband speech spectrum may be obtained by training process 10 multiplying the extended envelope with the extended excitation, as shown Fig. 4. This is again shown in example Fig. 5, depicting example spectrum 500 of a spectral envelope 502, an example excitation 504 and the synthesized spectrum 506. From this, it becomes clear that the envelope describes the spectral coarse structure while the excitation describes the spectral fine structure.
• the incoming narrow-band signal may be cut into overlapping windows, e.g., of 16 to 32ms duration. These windows may be separately analyzed by Fast Fourier Transform (FFT), e.g., a Short-time Fourier Transform (STFT) may be performed.
• FFT Fast Fourier Transform
• STFT Short-time Fourier Transform
• the bandwidth extended signal may be resynthesized by using an Inverse STFT (ISTFT) in combination with the overlap-and-add method.
• ISTFT Inverse STFT
• DNNs deep neural networks
• This is generally realized by training (e.g., via training process 10) a regression DNN to estimate wideband envelopes from given narrowband envelopes.
• MFCCs Mel Frequency Cepstral Coefficients
• the DNN is often fed with additional input features such as the first and second order derivative of MFCCs with respect to time (called delta and delta-delta features), the spectral centroid, zero-crossing rate, kurtosis, gradient index, noise-related frame energy, correlation coefficients and so on.
• FD denotes the frequency domain
• TD denotes the time domain
• HP denotes a high pass filter with a cut-off frequency of, e.g., 4kHz, such that the original 4kHz of the NB signal is preserved and only the signal above 4kHz is extended.
• Alternative implementations by training process 10 may perform the signal decomposition in the time domain using the LPC coefficients as a spectral envelope and the LPC residual as an excitation signal.
• the input features may be represented to the network in the nodes (e.g., nodes 700) of the input layer, shown in example Fig. 7. This may be followed by several hidden layers.
• the output of the network may be contained in the nodes of the output layer and may consist of the estimated wideband spectrum, possibly in a compressed form such as a Mel spectrum or MFCCs.
• the activations + 1 ⁇ ⁇ + F t a ⁇ + i , «l 0 f the nodes of the ( l +l)-st layer may be determined from the activations of the preceding layer.
• a weight matrix is a vector of biases and acl: is a non-linear activation function, such as a sigmoid 802, hyperbolic tangent (tanh) 804 or a rectified linear unit (Relu) 806 shown in example Fig. 8.
• acl is a non-linear activation function, such as a sigmoid 802, hyperbolic tangent (tanh) 804 or a rectified linear unit (Relu) 806 shown in example Fig. 8.
• Fig. 9 shows an example architecture 900 that again portrays how a particular activation + N may be determined from the activation vector ° of the preceding layer.
• the DNN training may be performed by training process 10 on a joint corpus of narrow-band and wideband speech.
• the narrow-band signal may be used by training process 10 for extracting the input features of the DNN.
• the corresponding wideband signal may be used by training process 10 for generating the target for the output layer of the DNN, e.g., the wideband spectral envelopes that the DNN should generate for the given input.
• Training a DNN may use a cost function, e.g., a measure between the desired target outputs and the outputs generated by the network, which is to be minimized during training by training process 10.
• a cost function e.g., a measure between the desired target outputs and the outputs generated by the network, which is to be minimized during training by training process 10.
• MSE mean squared error
• L-2 norm of the DNN weights e.g.,:
• [0074] is the L-2 norm over all weights matrices ⁇ ⁇ 1 ' ⁇ 2 ' It may be added to the cost function, e.g., in order to prevent too large weights in the training process. This is generally considered to improve the generalizability, e.g., the robustness of the trained network to unseen conditions and is part of standard training recipes.
• the actual training process of the DNN may be essentially a gradient descent algorithm. It may consist in first calculating the gradient of the cost function with respect to the weights and biases h 'V on he entire batch or mini-batch, and then taking a step into the opposite direction in order to reduce the cost, e.g.,:
• Atypical batch size may contain, e.g., a few seconds of speech data.
• the term f 1 denotes the step size. It has a major effect on the convergence speed and the performance of the trained network, and it may be determined automatically in modern DNN training toolkits, using, e.g., Adaptive Moment Estimation (ADAM).
• ADAM Adaptive Moment Estimation
• the gradient calculation may be accomplished with a back propagation method and it, in particular, may involve the calculation of the gradient with respect to the network outputs V U, e.g., the activations at the output layer of the network for a given input feature Xt , e.g.,:
• the gradient descent may be repeated until a specified stopping criterion has been fulfilled, such as no significant reduction of the cost C(y,y. W) the last M iterations on a validation data set that differs from the training data set.
• MSE as a cost function
• over-smoothing e.g., vowels and fricatives are extended in a similar fashion.
• fricatives are not generally extended strongly enough while vowels are extended too strongly, as shown in Bandwidth-expanded speech with MSE cost function 1002 and True wideband speech 1004 of example Fig. 10. This in particular may happen for data that differs from the training conditions, but it may also be observed on the training data, at least to some extent.
• training process 10 may transform 1200, by a computing device, a speech signal into a speech signal representation.
• Training process 10 may train 1202 a regression deep neural network with a cost function to minimize a mean squared error between actual values of the speech signal representation and estimated values of the speech signal representation, wherein the cost function may include one or more discriminative terms.
• Training process 10 may extend 1204 the bandwidth of the speech signal by extending the speech signal representation of the speech signal using the regression deep neural network trained with the cost function that includes the one or more discriminative terms.
• training process 10 may receive a speech signal, and transform 1200 the speech signal into a speech signal representation.
• the speech signal representation may be obtained by training process 10 decomposing 1208 the speech signal into a spectral envelope and an excitation signal, and wherein the spectral envelope may be extended using the regression deep neural network trained with the cost function. It will be appreciated that decomposing 1208 the speech signal may not be necessary (e.g., by directly estimating the entire complex wideband spectrum with a much larger DNN or convolutional neural network (CNN)).
• CNN convolutional neural network
• training process 10 may train 1202 a regression deep neural network with a cost function to minimize a mean squared error between actual values of the speech signal representation and estimated values of the speech signal representation, as well as extend 1204 bandwidth of the speech signal.
• training process 10 may overcome the above-noted over-smoothing problem by, e.g., improving the separation of different phoneme classes. This may be achieved by training process 10 adding one or more discriminative terms to the cost function in order to preserve differences between different phoneme classes, which may be most generally be formulated as, e.g.,:
• ⁇ denotes phoneme class labels corresponding to the spectra y ⁇ ⁇ yi . VN) 0 f the batch. More particularly, the labels may identify to which phoneme or phoneme class the spectra belong.
• the weight Y may be used by training process 10 to trade-off the discriminative measure versus the MSE, as well as the regression term.
• the one or more discriminative terms may preserve relations of statistics between different phoneme classes in the actual values of the speech signal representation and the estimated values of the speech signal representation.
• the different phoneme classes may include a fricative phoneme class and/or a vowel phoneme class. It will be appreciated that other phoneme classes may also be used with the present disclosure.
• the discriminative term(s) may include at least one of a fricative-to-vowel power ratio and a function thereof, and in some implementations, the relative statistic to be preserved by training process 10 may be the average high-band power ratio between fricatives and vowels (FVPR), e.g.,:
• training process 10 may use power ratios in different frequency bands of the spectrum, e.g., fricative-to-vowel power ratio calculated on the 4 frequency bands 4-5kHz, 5-6kHz, 6-7kHz, 7-8kHz, instead of using the broad band power ratio calculated from 4 to 8kHz (as discussed throughout).
• adding a term to the cost function that forces the DNN to match the variance or standard deviation of the true and estimated speech signal representations may also help with training a more discriminative network.
• a network trained with this term may tend to produce more artifacts than explicitly separating phoneme classes.
• the output features may need to be brought to the power spectral domain by training process 10 before calculating the discriminative measure.
• this may be achieved by, e.g., training process 10 multiplying the with the pseudo-inverse of the Discrete Cosine Transform
• training process 10 may take the inverse of
• training process 10 may reproduce 1206 an average power ratio at an output of the regression deep neural network.
• training process 10 adding this measure to the cost function may result in a joint optimization of the MSE and the discriminative term. If the weight Y is chosen appropriately, the network may be forced to approximately reproduce the true FVPR (at the output of the regression DNN) in addition to minimizing the mean squared error.
• This may be seen in the example plots 1302 and 1304 of example Fig. 13, which show the convergence throughout the training process of both PVPR(3/,L) ⁇ FVPR(y , L) as we
• the dashed curves show the cost measures for plain MSE training with regularization.
• the solid curves show the corresponding cost measures with the additional discriminative term.
• the dashed curve clearly shows that minimizing the MSE does not necessarily naturally minimize the distance between the FVPR of the DNN and that of true wideband speech.
• a bias and the power ratio may be systematically under-estimated.
• Training process 10 using the additional discriminative term preserves the FVPR (distance close to zero) but the MSE may converge to almost the same value as for plain MSE training. Consequently, the estimated wideband envelopes of fricatives and vowels may be better separated as well as the true wideband envelopes of the training data set.
• the cost function may preserve a power ratio between the different phoneme classes in the actual values of the speech signal representation and the estimated values of the speech signal representation, and in some implementations, the cost function may preserve the power ratio between the different phoneme classes using a weighted sum of K power ratio errors between the different phoneme classes. For example, training process 10 may generalize this to multiple phoneme classes, by, e.g., extending the distance measure to a weighted sum of ⁇ power ratio errors between different classes, e.g.,:
• [0095] denote the phoneme classes that are compared in the ⁇ -th ratio.
• the Uk are overestimation factors for the power ratios of the true wideband signals.
• ⁇ Disc(y yJ) ma y b e an y distance metric between statistics that relate different phoneme classes.
• training process 10 may use the present disclosure to learn to predict a value or a vector of values (similar to statistical regression). This contrasts with classification DNNs, which typically only learn class affiliations of the input feature vector (e.g., phoneme classes in the context of speech recognition).

Landscapes

• Engineering & Computer Science (AREA)
• Physics & Mathematics (AREA)
• Computational Linguistics (AREA)
• Health & Medical Sciences (AREA)
• Audiology, Speech & Language Pathology (AREA)
• Human Computer Interaction (AREA)
• Acoustics & Sound (AREA)
• Multimedia (AREA)
• Artificial Intelligence (AREA)
• Signal Processing (AREA)
• Theoretical Computer Science (AREA)
• Evolutionary Computation (AREA)
• Biophysics (AREA)
• General Engineering & Computer Science (AREA)
• Life Sciences & Earth Sciences (AREA)
• Data Mining & Analysis (AREA)
• General Health & Medical Sciences (AREA)
• Molecular Biology (AREA)
• Computing Systems (AREA)
• Biomedical Technology (AREA)
• General Physics & Mathematics (AREA)
• Mathematical Physics (AREA)
• Software Systems (AREA)
• Computer Vision & Pattern Recognition (AREA)
• Spectroscopy & Molecular Physics (AREA)
• Quality & Reliability (AREA)
• Telephone Function (AREA)
• Mobile Radio Communication Systems (AREA)

Priority Applications (2)

Applications Claiming Priority (2)

Publications (1)

Family

ID=68238175

Family Applications (1)

Country Status (4)

Families Citing this family (5)

Citations (3)

Family Cites Families (9)

• 2018
  • 2018-04-23 US US15/959,606 patent/US10650806B2/en active Active
• 2019
  • 2019-04-23 CN CN201980028119.0A patent/CN112088385A/zh active Pending
  • 2019-04-23 EP EP19792022.6A patent/EP3785189B1/en active Active
  • 2019-04-23 WO PCT/US2019/028742 patent/WO2019209841A1/en unknown

Patent Citations (3)

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events